Global high resolution wind speed statistics from satellite lidar measurement below the surface to 8 km above with 30 m vertical resolution. Backscatter data at coarser resolution are Yongxiang Hu Carl Weimer collected below and above that vertical altitude range Climate Science Branch Ball Aerospace Corp. (Figure 1). NASA LaRC Boulder, CO Hampton, Virginia 23681 [email protected] [email protected] Abstract— The lidar on the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) mission provides accurate ocean surface backscatter intensity at 1064nm wavelength. The measurements can be used for estimating ocean surface wind speed at CALIPSO’s 70 m lidar footprint. We are developing the high spatial resolution ocean surface wind speed data product for all the CALIPSO measurements, starting from June 2006 and ending at the end of CALIPSO mission (likely 2012). Wind Speed probability distributions will be studied based on the high spatial resolution data and compared with QuikScat. Scale and shape parameters of the Weibull wind distributions are derived from the wind speed product, and their global and regional statistics, together with the CALIPSO wind speed product, will be made available to the research community. (Abstract) Keywords-component; formatting; style; styling; insert (key words) 1. CALIPSO lidar measurements The Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) mission is a satellite mission being developed within the framework of collaboration between NASA and the French space agency, CNES. The CALIPSO mission provides unique measurements to improve our understanding of the role of aerosols and clouds in the Earth’s climate system. One of the CALIPSO mission’s payload is a two- wavelength polarization-sensitive lidar. A diode-pumped Nd:YAG laser produces linearly-polarized pulses of light at 1064 nm and 532 nm. The atmospheric return is collected by a 1-meter telescope which feeds a three channel receiver Figure 1 CALIPSO lidar measurements and vertical range measuring the backscattered intensity at 1064 nm and the resolution. The laser beam size is 70 m. two orthogonal polarization components at 532 nm (parallel and perpendicular to the polarization plane of the transmitted beam). Each laser produces 110 mJ of energy at 2. Mean square slope estimates from CALIPSO ocean each of the two wavelengths at a pulse repetition rate of 20.2 Hz. Only one laser is operated at a time. Beam surface lidar backscatter expanders reduce the angular divergence of the laser beam The parallel polarization component of the CALIPSO to produce a beam diameter of 70 meters at the Earth’s lidar backscatter from ocean surfaces is primarily a result of surface. The fundamental sampling resolution of the lidar is specular reflection. When a laser beam hits a water surface 30 meters vertical and 333 meters horizontal, determined by at near normal incidence, about 2% of the laser energy is the receiver electrical bandwidth and the laser pulse reflected and the rest of the energy goes into the water. For repetition rate. Backscatter data are acquired from 0.5 km weak winds, the water surface is smooth and specular reflection is a narrow beam with little divergence. As wind properties, such as skewness and peakedness, can be speed increases, the surface roughens and the divergence ignored. The independent absolute calibration of CALIPSO angle of the same 2% of specularly reflected energy lidar measurements enables an accurate assessment of the increases. Thus, lidar backscatter intensity from the ocean relations by comparing the AMSR-E wind speeds with the surface decreases as wind speeds increase. Simple relations variance information derived from lidar backscatter (Figure [e.g., Cox and Munk, 1954; Wu, 1990; Hu et al., 2008] 3). between wind speed and ocean surface roughness (wave By definition, the CALIPSO lidar sea surface slope variance) can be applied to estimate ocean surface backscatter, γ, is wind speed from 1064nm lidar clear sky ocean measurements. It has been over half a century since Cox and Munk (1954) introduced the Gaussian distribution relation For near-normal incidence, the Fresnel reflectance, ρ, between sea surface wind and the slopes of wind driven is 0.0209 for sea water at 532 nm and 0.0193 at 1064nm. waves. A Gaussian distribution has maximum information For 0.3 degree off-nadir lidar pointing, tan2θ is very close to entropy [P(si)logP(si), where P(si) is the probability of slope 0 and cos2θ is very close to 0. Thus, at 1064nm wavelength, si] and thus is the most probable state if we can consider the wind driven slopes (si) as many independent and . identicallydistributed random variables [central limit theorem]. In the same study, Cox and Munk (1954) also So, the ocean surface lidar backscatter from CALIPSO, γ, is suggested a linear relationship between wind speed at a direct measurement of mean square slope, <s2> (Figure 2). 12.5m above sea surface and the mean square slope, based on measurements of the bi-directional sea surface reflectance pattern of reflected sunlight. Using laboratory measurements, Wu (1990) revised the relation between wind speed and slope variance to two log-linear relations. When the wind speed is less than 7 m/s, capillary waves, resulting from a balance between atmospheric wind friction and water surface tension, is the predominant component of wind driven waves. For wind speed exceeding 7 m/s, the surface becomes rougher and the predominant wavelengths grow to centimeter scale while surface tension weakens and gravity becomes more important in terms of restoring surface smoothness (i.e., gravity-capillary waves). The wind speed – wave slope variance relation also varies with sea surface state and meteorological conditions (Shaw and Figure 2. Mean square slope estimated from CALIPSO 1064nm Churnside, 1997). ocean surface backscatter (January 2008). Using collocated TRMM sea surface radar cross sections and wind speeds from microwave radiometer, On average, the signal-noise-ratio (SNR) for a single lidar Freilich and Vanhoff (2003) analyzed the wind speed – shot ocean surface backscatter measurement (70 m laser mean square slope relations on a global scale and spot) is around 100 at nighttime (better at lower wind speed demonstrated that at lower wind speed (U<10 m/s), a log- and worse at higher wind speed). After accounting for linear relation agrees well with the observations, while at uncertainties such as aerosol optical depth calculations and larger wind speed (5 m/s<U<19 m/s), both the linear Cox- calibration, the instantaneous error of the mean square slope Munk relation and log-linear Wu relation are within the estimates at 70 m lidar footprint from CALIPSO uncertainty of observation. The wave slope variances measurements is around 5-10% (which translates to derived from microwave data are slightly different from instantaneous wind speed uncertainty of around 1 m/s). those derived from visible and infrared measurements since they cover different wave-number ranges of wind-generated 3. How can these high spatial resolution measurements waves (Liu et al., 2000). A global analysis of wind speed – mean square slope relation for waves that fall within the compliment QuickScat lidar backscatter sensitivity range can be performed by Due to the limitation of coverage (nadir viewing only), the comparing sea surface backscatter of the CALIPSO lidar use of the global CALIPSO wind speed observations is with the collocated wind speed measurements of AMSR-E limited, comparing with QuikScat data. But the high spatial on the Aqua spacecraft. As the space-based lidar onboard resolution (70 m) wind speed statistics from the lidar the CALIPSO satellite only measures sea surface measurements do bring important information that can backscatter at a 0.3 degree off-nadir angle, directional compliment QuikScat. With the70 m spatial sampling of the wind speed, CALIPSO Figure 5 shows that nearly almost everywhere, the captures more wind gusts and provide global statistics of standard deviations of the monthly mean winds from that. As a result, the wind speed distribution is broader than CALIPSO measurements are much higher than the standard measurements with larger footprint sizes (e.g., QuikScat and deviation of AMSR-E winds. AMSR-E). As demonstrated in Figure 4, the CALIPSO wind speed (70 m)distribution has more low and high winds than AMSR-E (40 km), while the mean wind speeds of CALIPSO and AMSR-E are very close (Hu et al., 2008). Figure 5 Standard deviations (m/s) of Jan 2008 wind speed from Figure 3. Frequency of occurrence (color bar: log scale) of the AMSR‐E (upper) and CALIPSO (lower). AMSR‐E wind speed – CALIPSO mean square wave slope relation. The vertical turbulent transport velocities of heat and Using The wind speed – wave slope relations derived from gases are sensitive not only to the mean surface wind speed, the collocated CALIPSO and AMSR-E measurements as but also the detailed wind speed probability distributions illustrated in figure 3, wind speed at CALIPSO laser sot size (e.g., Winninkhof et al., 1992). (70 m) resolution can be derived (Hu et al., 2008). Wind speed probability distribution is normally represented by Weibull distribution (e.g., Justus et al., 1978; Monahan, 2006), . The parameters a and b denote the scale and shape parameters of the distribution (Justus et al., 1978), The scale parameters, a, for both AMSR-E and CALIPSO are similar (upper panel of Fig 6). But the shape parameters, b, are very different (lower panels of Fig 6). Due to AMSR-E’s narrower wind speed distribution, the b Figure 4 Global ocean surface wind speed distribution of Jan. values of AMSR-E are much higher than the CALIPSO b 2008 from CALIPSO (blue) and AMSR‐E (red) collocated values. measurements. Figure 7. the relation between a and b of Weibull distributions from AMSR‐E (blue) and CALIPSO (red).